Despite the rarity of their impacting Earth, long-period comets are recognized as potentially the most devastating impact threat to our planet (i.e., an extinction-level event) due to their large size and fast travel speeds of up to 70 km/s. Evidence indicates that the impact of a comet or asteroid with a diameter of about 10 km was responsible for the mass extinction of most species of dinosaurs and marine life. However, any new comet on an impact trajectory with Earth would likely only be discovered 6–12 months before impact, when it becomes visible as the Sun’s heat and wind start sublimating its icy surface and ejecting rocky debris.
Siddha Ganju explains how the FDL lab at NASA uses artificial intelligence to improve and automate the identification of meteors above human level performance using meteor shower images and recover known meteor shower streams and characterize previously unknown meteor showers using orbital data—research aimed at providing more warning time for long-period comet impacts.
The Cameras for Allsky Meteor Surveillance (or CAMS) is a network of low-light video cameras, established by the SETI Institute and NASA in different locations across the globe, that monitors the sky to detect meteors. Until now, processing the images collected by CAMS has required time-consuming human input. On an average night, an astronomer receives around 500 detections per camera, consisting of images and light intensity curves (a sequence of measurements of how light intensity changes as detected objects move in the sky). This totals 8,000 observations, with 16 cameras per site, most of which turn out to be false detections, such as planes, birds, clouds, etc. (Only about 15% are actually meteors.)
Sorting through these every night is not scalable. To address this challenge, the process has been automated using deep learning—the first time that deep learning techniques have been applied to this endeavor. Deep learning algorithms such as temporal object localization based on a combination of CNNs and LSTMs are used to recognize meteors among false positives and are trained in such a way that they are able to triangulate the meteor trajectory in Earth’s atmosphere, its entry speed, and the pre-impact orbit in space through combining different camera perspectives for the same meteor.
The goal was to add years of extra warning time by providing comet searchers directions on where to look for comets when they are still far out. The task was particularly suited for machine learning approaches because of the large scale of data, the need for automation (as all the meteor shower surveys from around the globe need to be integrated without human intervention), and the need to operate for a long period of time. Siddha discusses the machine learning and deep learning pipelines aimed at fulfilling the above conditions.
Siddha Ganju is a self-driving architect at NVIDIA. She was featured on the Forbes 30 under 30 list, and she guides teams at NASA as an AI domain expert and is a featured jury member in several informational tech competitions. Previously, she developed deep learning models for resource-constrained edge devices at DeepVision. She earned her degree from Carnegie Mellon University, and her work ranges from visual question answering to generative adversarial networks to gathering insights from CERN’s petabyte-scale data and has been published at top-tier conferences including CVPR and NeurIPS.
Comments on this page are now closed.
©2018, O'Reilly Media, Inc. • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • email@example.com